[SPMD] Add apply_backward_optimization_barrier #6097
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
alanwaketan
commented
Dec 11, 2023
test/spmd/test_xla_sharding.py
Outdated
| # The first layer won't have gradients in the hook. Not sure why. | ||
| xs.xla_sharding.apply_backward_optimization_barrier(model.fc2) | ||
|
|
||
| # optimizer.zero_grad() |
There was a problem hiding this comment.
I should remove this. oops...
|
Thanks Jon for approving the pull request. |
JackCaoG
approved these changes
Dec 11, 2023
Summary: This pull request adds a new API to xla_sharding.py called apply_backward_optimization_barrier where registers a full backward hook that apply an optimization barrier to the given module. This API will prevent the XLA compiler from fusing the module's backward pass with others. And It's useful to prevent gigantic buffers being allocated to synchronize the gradients. Test Plan: python test/spmd/test_xla_sharding.py -v -k test_backward_optimization_barrier
alanwaketan
force-pushed
the
alanwaketan/opt-barrier
branch
from
23d1272 to
cb3bad6
Compare
chunnienc
pushed a commit
to chunnienc/xla
that referenced
this pull request
Dec 14, 2023
Summary: This pull request adds a new API to xla_sharding.py called apply_backward_optimization_barrier where registers a full backward hook that apply an optimization barrier to the given module. This API will prevent the XLA compiler from fusing the module's backward pass with others. And It's useful to prevent gigantic buffers being allocated to synchronize the gradients. It's also used in pytorch-tpu/transformers#50. Test Plan: python test/spmd/test_xla_sharding.py -v -k test_backward_optimization_barrier
golechwierowicz
pushed a commit
that referenced
this pull request
Jan 12, 2024
Summary: This pull request adds a new API to xla_sharding.py called apply_backward_optimization_barrier where registers a full backward hook that apply an optimization barrier to the given module. This API will prevent the XLA compiler from fusing the module's backward pass with others. And It's useful to prevent gigantic buffers being allocated to synchronize the gradients. It's also used in pytorch-tpu/transformers#50. Test Plan: python test/spmd/test_xla_sharding.py -v -k test_backward_optimization_barrier
bhavya01
pushed a commit
that referenced
this pull request
Apr 22, 2024
Summary: This pull request adds a new API to xla_sharding.py called apply_backward_optimization_barrier where registers a full backward hook that apply an optimization barrier to the given module. This API will prevent the XLA compiler from fusing the module's backward pass with others. And It's useful to prevent gigantic buffers being allocated to synchronize the gradients. It's also used in pytorch-tpu/transformers#50. Test Plan: python test/spmd/test_xla_sharding.py -v -k test_backward_optimization_barrier
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
This pull request adds a new API to xla_sharding.py called apply_backward_optimization_barrier where registers a full backward hook that apply an optimization barrier to the given module. This API will prevent the XLA compiler from fusing the module's backward pass with others. And It's useful to prevent gigantic buffers being allocated to synchronize the gradients.
It's also used in pytorch-tpu/transformers#50.
Test Plan:
python test/spmd/test_xla_sharding.py -v -k test_backward_optimization_barrier